bf-update-conventions
Pass
Audited by Gen Agent Trust Hub on Feb 25, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses system commands to interact with local repository history.
- Evidence: The instructions specify using
git logto extract patterns from commit messages andgit committo finalize updates to documentation files. These are appropriate and expected operations for this skill's context. - [PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection (Category 8) due to its data processing model.
- Ingestion points: It reads untrusted textual data from
docs/archive/{TICKET}/reviews/*.mdandgit logoutput. - Boundary markers: The process lacks explicit delimiters or specific instructions for the agent to ignore any embedded commands within the review data.
- Capability inventory: The skill is authorized to modify sensitive instruction-bearing files, including
docs/conventions.md(used as a rule source) andCLAUDE.md. - Sanitization: There is no specified logic to sanitize or validate the patterns extracted from the external logs before they are appended to the project's convention documents.
Audit Metadata